Back to all posts
Projects & Retrofits7 min read·November 15, 2026

How Do You Map a Production Process Before Automating It?

TB
ThynkBlox Automation Team
Automation Engineering

To map a business process before automating it, draw the process as it actually runs today on one page. Use four swimlanes: operator, machine or PLC, MES and ERP. Show every step, every handoff between lanes, the data that moves at each handoff, and every decision point. Then walk the line with the people who run it so you capture the manual workarounds and undocumented steps that never appear in a procedure. That map is the input to your specification. It is not a design.

Why map the process before you automate it?

Automation copies whatever logic you give it. If the logic you hand over is the official procedure, and the line really runs on a clipboard, a phone call and an operator's memory, the new system will be built for a process that does not exist.

A map gives you three things before any money goes on hardware or software:

  • A shared picture that operators, maintenance, planning and the integrator all agree is accurate.
  • A list of handoffs. Most automation problems sit here: data retyped, approvals given verbally, files carried between systems.
  • A scope boundary. You can point at the map and say which steps the project changes and which it leaves alone.

The map then feeds the functional design specification, which describes how the future system should behave.

What goes on a one-page swimlane map?

Use one horizontal lane per actor. For most production processes four lanes are enough:

LaneWhat belongs in itTypical examples
OperatorHuman actions and judgementsLoading material, confirming a changeover, inspecting a sample
Machine or PLCAutomatic actions and signalsStart and stop, counts, fault codes, interlocks
MESShop-floor records and trackingOrder start, batch identity, quality results, downtime
ERPBusiness recordsProduction order, material issue, finished goods receipt

If your site has no MES, keep the lane anyway and write what fills that role today: a spreadsheet, a paper log or a shift report. An empty lane is information.

Draw four kinds of element:

  1. Steps. A box with a verb and an object, such as "Print case label". One action per box.
  2. Handoffs. An arrow that crosses a lane boundary. Label it with how it happens: signal, screen entry, paper, verbal, file.
  3. Data. Next to each handoff, write the items that move: order number, product code, quantity, batch, timestamp.
  4. Decisions. A diamond with a yes or no question, such as "Label matches order?" Show both exits.

Keep it on one page. If it will not fit, you are mapping at too fine a level. Put the detail in a separate sheet for that step and keep the overview readable.

How do you capture the process as it really runs?

Start with whatever documents exist, then go to the line. Watch at least one full cycle, including a product changeover and a shift handover if they occur. Ask the operator to talk through what they are doing and why.

Useful questions while you watch:

  • Where did that number come from?
  • What would you do if this screen were down?
  • Who do you tell when this happens, and how?
  • Is there anything you check that is not written down?
  • What do night shift do differently?

Mark each step on the map as observed, described or assumed. Assumed steps need confirming before the map is accepted. Where control drawings or documentation are missing, combine this walk with a survey of a plant with missing control drawings so the PLC lane reflects the real signals.

How do you find manual handoffs, workarounds and undocumented steps?

Run through this checklist against the draft map:

  • Retyped data. Any value read from one screen or document and entered into another.
  • Paper in the loop. Printed orders, tally sheets, labels written by hand, sign-off forms.
  • Verbal handoffs. Radio calls, phone calls or shouts that start or stop a step.
  • Personal files. Spreadsheets or notes kept by one person that others depend on.
  • Bypasses. Jumpered interlocks, forced bits, override keys or skipped screens used to keep running.
  • Repeat checks. The same thing checked by two people or two systems because neither is trusted.
  • Waiting. Steps that stop until someone arrives, approves or finds information.
  • Exceptions. What happens on rework, a short order, a split batch or a rejected pallet.
  • End of shift and end of order. Counts reconciled, adjustments booked, material returned.
  • Single points of knowledge. Steps only one person knows how to do.

Treat any bypass of a safety function as a safety matter, not a mapping note. Report it immediately and involve a qualified assessor before the project decides anything about it.

What does a worked packing-line example look like?

Illustrative example: a line fills cartons, packs them into cases and palletises them. The steps and handoffs below are invented to show the method.

StepLaneHandoff and methodDataFinding
Release production orderERPTo supervisor, printed sheetOrder number, product, quantityPaper handoff
Set up line for productOperatorTo PLC, recipe chosen on HMIProduct codeOperator selects recipe from memory of the sheet
Start orderOperatorTo MES, typed inOrder numberOrder number retyped
Fill and count cartonsPLCNone recordedCarton countCount not sent to MES
Print case labelsOperatorLabel printer, manual editBatch, best-beforeBatch typed by hand each shift
Decision: label correct?OperatorVisual checkNoneNot documented in procedure
Record outputOperatorTally sheet, then MES at shift endCases per palletPaper, delayed entry
Confirm order to ERPSupervisorTyped from MES report next morningGood quantity, scrapRetyped, delayed

From this map the team sees five manual handoffs, one undocumented decision and one count that exists in the PLC but never reaches a record. Those become the candidate scope: pass the order from ERP to MES and PLC electronically, take the count from the PLC, and generate the label data from the order. The design of that order exchange is covered in exchanging production orders between ERP and the shop floor.

Notice what the map does not decide. It does not choose software or say how the label check should work in future. It shows where the effort and risk sit.

What are the common mistakes?

Mapping the procedure instead of the practice. The written procedure is a starting point. Confirm every step on the floor.

Mapping only the happy path. Rework, rejects and short orders are where manual steps hide.

Leaving out the data. A handoff arrow without the data it carries cannot be turned into a requirement.

Designing while mapping. Record the current state first. Put improvement ideas in a separate list.

Skipping the people who run it. A map drawn in an office and never reviewed by operators will be wrong in ways nobody notices until commissioning.

Treating the map as finished. Date it, give it an owner and update it when the process changes.

When does this approach not apply?

A full four-lane map is more than you need for a single standalone machine with no business data, such as a replacement conveyor drive. A simple sequence description is enough there. It is also not a substitute for a hazard study: the map describes flow and information, not risk.

Frequently asked questions

How long should mapping one process take?

It depends on the process complexity and how many shifts and product variants exist. Plan time to observe at least one complete cycle, a changeover and a shift handover, then a review session where operators, supervisors and maintenance confirm the draft before it is used.

Who should own the process map?

Give ownership to someone in operations who runs or supervises the process, not to the integrator. The integrator can draw it, but the plant must confirm it is accurate and keep it current, because it later becomes the reference for scope disputes and acceptance tests.

Do we need special software to draw the map?

No. A whiteboard, paper sheets or any drawing tool will do. What matters is that lanes, steps, handoffs, data and decisions are clear and that the result fits one readable page. Agree a single format so later revisions can be compared easily.

What do we do with the map once it is agreed?

Use it to set scope, list requirements for the functional design specification and identify interfaces between systems. Keep the current-state version as a baseline, then draw a future-state map so everyone can see exactly which steps and handoffs the project changes.

Next step

Want to apply this to your plant?

Every project starts with a fixed-fee assessment, and you can stop after the report.